binsequencecommand.cpp

   1 /*
   2  *  binsequencecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 4/3/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amhers. All rights reserved.
   7  *
   8  */
   9
  10 #include "binsequencecommand.h"
  11
  12 //**********************************************************************************************************************
  13 BinSeqCommand::BinSeqCommand(string option){
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16                 abort = false;
  17                 allLines = 1;
  18                 labels.clear();
  19
  20                 //allow user to run help
  21                 if(option == "help") { help(); abort = true; }
  22
  23                 else {
  24                         //valid paramters for this command
  25                         string AlignArray[] =  {"fasta","label","name", "group","outputdir","inputdir"};
  26                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  27
  28                         OptionParser parser(option);
  29                         map<string, string> parameters = parser.getParameters();
  30
  31                         ValidParameters validParameter;
  32                         map<string, string>::iterator it;
  33
  34                         //check to make sure all parameters are valid for command
  35                         for (it = parameters.begin(); it != parameters.end(); it++) {
  36                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  37                         }
  38
  39                         //if the user changes the output directory command factory will send this info to us in the output parameter
  40                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  41                                 outputDir = "";
  42                                 outputDir += hasPath(globaldata->getListFile()); //if user entered a file with a path then preserve it
  43                         }
  44
  45
  46                         //make sure the user has already run the read.otu command
  47                         if (globaldata->getListFile() == "") {
  48                                 mothurOut("You must read a listfile before running the bin.seqs command.");
  49                                 mothurOutEndLine();
  50                                 abort = true;
  51                         }
  52
  53                         //if the user changes the input directory command factory will send this info to us in the output parameter
  54                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  55                         if (inputDir == "not found"){   inputDir = "";          }
  56                         else {
  57                                 string path;
  58                                 it = parameters.find("fasta");
  59                                 //user has given a template file
  60                                 if(it != parameters.end()){
  61                                         path = hasPath(it->second);
  62                                         //if the user has not given a path then, add inputdir. else leave path alone.
  63                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  64                                 }
  65
  66                                 it = parameters.find("name");
  67                                 //user has given a template file
  68                                 if(it != parameters.end()){
  69                                         path = hasPath(it->second);
  70                                         //if the user has not given a path then, add inputdir. else leave path alone.
  71                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
  72                                 }
  73
  74                                 it = parameters.find("group");
  75                                 //user has given a template file
  76                                 if(it != parameters.end()){
  77                                         path = hasPath(it->second);
  78                                         //if the user has not given a path then, add inputdir. else leave path alone.
  79                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
  80                                 }
  81                         }
  82
  83
  84                         //check for required parameters
  85                         fastafile = validParameter.validFile(parameters, "fasta", true);
  86                         if (fastafile == "not found") { mothurOut("fasta is a required parameter for the bin.seqs command.");  mothurOutEndLine(); abort = true; }
  87                         else if (fastafile == "not open") { abort = true; }
  88
  89                         //check for optional parameter and set defaults
  90                         // ...at some point should added some additional type checking...
  91
  92                         label = validParameter.validFile(parameters, "label", false);
  93                         if (label == "not found") { label = ""; }
  94                         else {
  95                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  96                                 else { allLines = 1;  }
  97                         }
  98
  99                         //if the user has not specified any labels use the ones from read.otu
 100                         if (label == "") {
 101                                 allLines = globaldata->allLines;
 102                                 labels = globaldata->labels;
 103                         }
 104
 105                         namesfile = validParameter.validFile(parameters, "name", true);
 106                         if (namesfile == "not open") { abort = true; }
 107                         else if (namesfile == "not found") { namesfile = ""; }
 108
 109                         groupfile = validParameter.validFile(parameters, "group", true);
 110                         if (groupfile == "not open") { abort = true; }
 111                         else if (groupfile == "not found") { groupfile = ""; }
 112
 113                         if (abort == false) {
 114 //                              openInputFile(fastafile, in);
 115                                 fasta = new FastaMap();
 116                                 if (groupfile != "") {
 117                                         groupMap = new GroupMap(groupfile);
 118
 119                                         int error = groupMap->readMap();
 120                                         if (error == 1) { delete groupMap; abort = true; }
 121                                 }
 122                         }
 123
 124                 }
 125         }
 126         catch(exception& e) {
 127                 errorOut(e, "BinSeqCommand", "BinSeqCommand");
 128                 exit(1);
 129         }
 130 }
 131 //**********************************************************************************************************************
 132
 133 void BinSeqCommand::help(){
 134         try {
 135                 mothurOut("The bin.seqs command can only be executed after a successful read.otu command of a listfile.\n");
 136                 mothurOut("The bin.seqs command parameters are fasta, name, label and group.  The fasta parameter is required.\n");
 137                 mothurOut("The label parameter allows you to select what distance levels you would like a output files created for, and are separated by dashes.\n");
 138                 mothurOut("The bin.seqs command should be in the following format: bin.seqs(fasta=yourFastaFile, name=yourNamesFile, group=yourGroupFile, label=yourLabels).\n");
 139                 mothurOut("Example bin.seqs(fasta=amazon.fasta, group=amazon.groups, name=amazon.names).\n");
 140                 mothurOut("The default value for label is all lines in your inputfile.\n");
 141                 mothurOut("The bin.seqs command outputs a .fasta file for each distance you specify appending the OTU number to each name.\n");
 142                 mothurOut("If you provide a groupfile, then it also appends the sequences group to the name.\n");
 143                 mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 144         }
 145         catch(exception& e) {
 146                 errorOut(e, "BinSeqCommand", "help");
 147                 exit(1);
 148         }
 149 }
 150
 151 //**********************************************************************************************************************
 152
 153 BinSeqCommand::~BinSeqCommand(){
 154         //made new in execute
 155         if (abort == false) {
 156                 delete input;  globaldata->ginput = NULL;
 157                 delete read;
 158                 globaldata->gListVector = NULL;
 159                 delete fasta;
 160                 if (groupfile != "") {  delete groupMap;  globaldata->gGroupmap = NULL; }
 161         }
 162 }
 163
 164 //**********************************************************************************************************************
 165
 166 int BinSeqCommand::execute(){
 167         try {
 168                 if (abort == true) {    return 0;       }
 169
 170                 int error = 0;
 171
 172                 //read fastafile
 173                 fasta->readFastaFile(fastafile);
 174
 175
 176                 //set format to list so input can get listvector
 177 //              globaldata->setFormat("list");
 178
 179                 //if user gave a namesfile then use it
 180                 if (namesfile != "") {
 181                         readNamesFile();
 182                 }
 183
 184                 //read list file
 185                 read = new ReadOTUFile(globaldata->getListFile());
 186                 read->read(&*globaldata);
 187
 188                 input = globaldata->ginput;
 189                 list = globaldata->gListVector;
 190                 string lastLabel = list->getLabel();
 191
 192                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 193                 set<string> processedLabels;
 194                 set<string> userLabels = labels;
 195
 196
 197                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 198
 199                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 200
 201                                 error = process(list);
 202                                 if (error == 1) { return 0; }
 203
 204                                 processedLabels.insert(list->getLabel());
 205                                 userLabels.erase(list->getLabel());
 206                         }
 207
 208                         if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 209                                 string saveLabel = list->getLabel();
 210
 211                                 delete list;
 212                                 list = input->getListVector(lastLabel);
 213
 214                                 error = process(list);
 215                                 if (error == 1) { return 0; }
 216
 217                                 processedLabels.insert(list->getLabel());
 218                                 userLabels.erase(list->getLabel());
 219
 220                                 //restore real lastlabel to save below
 221                                 list->setLabel(saveLabel);
 222                         }
 223
 224                         lastLabel = list->getLabel();
 225
 226                         delete list;
 227                         list = input->getListVector();
 228                 }
 229
 230
 231                 //output error messages about any remaining user labels
 232                 set<string>::iterator it;
 233                 bool needToRun = false;
 234                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 235                         mothurOut("Your file does not include the label " + *it);
 236                         if (processedLabels.count(lastLabel) != 1) {
 237                                 mothurOut(". I will use " + lastLabel + "."); mothurOutEndLine();
 238                                 needToRun = true;
 239                         }else {
 240                                 mothurOut(". Please refer to " + lastLabel + ".");  mothurOutEndLine();
 241                         }
 242                 }
 243
 244                 //run last label if you need to
 245                 if (needToRun == true)  {
 246                         if (list != NULL) {             delete list;    }
 247                         list = input->getListVector(lastLabel);
 248
 249                         error = process(list);
 250                         if (error == 1) { return 0; }
 251
 252                         delete list;
 253                 }
 254
 255                 return 0;
 256         }
 257         catch(exception& e) {
 258                 errorOut(e, "BinSeqCommand", "execute");
 259                 exit(1);
 260         }
 261 }
 262
 263 //**********************************************************************************************************************
 264 void BinSeqCommand::readNamesFile() {
 265         try {
 266                 vector<string> dupNames;
 267                 openInputFile(namesfile, inNames);
 268
 269                 string name, names, sequence;
 270
 271                 while(inNames){
 272                         inNames >> name;                        //read from first column  A
 273                         inNames >> names;               //read from second column  A,B,C,D
 274
 275                         dupNames.clear();
 276
 277                         //parse names into vector
 278                         splitAtComma(names, dupNames);
 279
 280                         //store names in fasta map
 281                         sequence = fasta->getSequence(name);
 282                         for (int i = 0; i < dupNames.size(); i++) {
 283                                 fasta->push_back(dupNames[i], sequence);
 284                         }
 285
 286                         gobble(inNames);
 287                 }
 288                 inNames.close();
 289
 290         }
 291         catch(exception& e) {
 292                 errorOut(e, "BinSeqCommand", "readNamesFile");
 293                 exit(1);
 294         }
 295 }
 296 //**********************************************************************************************************************
 297 //return 1 if error, 0 otherwise
 298 int BinSeqCommand::process(ListVector* list) {
 299         try {
 300                                 string binnames, name, sequence;
 301
 302                                 string outputFileName = outputDir + getRootName(getSimpleName(globaldata->getListFile())) + list->getLabel() + ".fasta";
 303                                 openOutputFile(outputFileName, out);
 304
 305                                 mothurOut(list->getLabel()); mothurOutEndLine();
 306
 307                                 //for each bin in the list vector
 308                                 for (int i = 0; i < list->size(); i++) {
 309
 310                                         binnames = list->get(i);
 311                                         while (binnames.find_first_of(',') != -1) {
 312                                                 name = binnames.substr(0,binnames.find_first_of(','));
 313                                                 binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 314
 315                                                 //do work for that name
 316                                                 sequence = fasta->getSequence(name);
 317                                                 if (sequence != "not found") {
 318                                                         //if you don't have groups
 319                                                         if (groupfile == "") {
 320                                                                 name = name + "|" + toString(i+1);
 321                                                                 out << ">" << name << endl;
 322                                                                 out << sequence << endl;
 323                                                         }else {//if you do have groups
 324                                                                 string group = groupMap->getGroup(name);
 325                                                                 if (group == "not found") {
 326                                                                         mothurOut(name + " is missing from your group file. Please correct. ");  mothurOutEndLine();
 327                                                                         remove(outputFileName.c_str());
 328                                                                         return 1;
 329                                                                 }else{
 330                                                                         name = name + "|" + group + "|" + toString(i+1);
 331                                                                         out << ">" << name << endl;
 332                                                                         out << sequence << endl;
 333                                                                 }
 334                                                         }
 335                                                 }else {
 336                                                         mothurOut(name + " is missing from your fasta or name file. Please correct. "); mothurOutEndLine();
 337                                                         remove(outputFileName.c_str());
 338                                                         return 1;
 339                                                 }
 340
 341                                         }
 342
 343                                         //get last name
 344                                         sequence = fasta->getSequence(binnames);
 345                                         if (sequence != "not found") {
 346                                                 //if you don't have groups
 347                                                 if (groupfile == "") {
 348                                                         binnames = binnames + "|" + toString(i+1);
 349                                                         out << ">" << binnames << endl;
 350                                                         out << sequence << endl;
 351                                                 }else {//if you do have groups
 352                                                         string group = groupMap->getGroup(binnames);
 353                                                         if (group == "not found") {
 354                                                                 mothurOut(binnames + " is missing from your group file. Please correct. "); mothurOutEndLine();
 355                                                                 remove(outputFileName.c_str());
 356                                                                 return 1;
 357                                                         }else{
 358                                                                 binnames = binnames + "|" + group + "|" + toString(i+1);
 359                                                                 out << ">" << binnames << endl;
 360                                                                 out << sequence << endl;
 361                                                         }
 362                                                 }
 363                                         }else {
 364                                                 mothurOut(binnames + " is missing from your fasta or name file. Please correct. "); mothurOutEndLine();
 365                                                 remove(outputFileName.c_str());
 366                                                 return 1;
 367                                         }
 368                                 }
 369
 370                                 out.close();
 371                                 return 0;
 372
 373         }
 374         catch(exception& e) {
 375                 errorOut(e, "BinSeqCommand", "process");
 376                 exit(1);
 377         }
 378 }
 379 //**********************************************************************************************************************
 380
 381